Real-Time Model-Based Inversion in Cross-Sectional Optoacoustic Tomography.

Area of Science:

• Biomedical Imaging
• Acoustic Physics

Background:

• Analytical inversion schemes are standard for optoacoustic tomography (OT) image reconstruction due to speed and low memory use.
• Accurate quantitative imaging and artifact reduction necessitate advanced inversion methods incorporating experimental factors.
• Model-based inversion in OT typically involves computationally intensive, memory-demanding algorithms, limiting real-time applications.

Purpose of the Study:

• To develop an efficient model-based reconstruction method for 2D optoacoustic tomography.
• To enable real-time image rendering in optoacoustic tomography using graphics processing units (GPUs).

Main Methods:

• A novel discretization procedure for efficient model-based reconstructions in 2D optoacoustic tomography.
• Parallel implementation on a graphics processing unit (GPU) with reduced numerical complexity.
• On-the-fly calculation of the model matrix within each inversion iteration.

Main Results:

• Achieved efficient model-based reconstructions with relatively low numerical complexity.
• Enabled parallel implementation on GPUs for optoacoustic tomography.
• Attained imaging frame rates exceeding 10 Hz, facilitating real-time image rendering.

Conclusions:

• The new discretization procedure significantly enhances the efficiency of model-based optoacoustic tomography.
• Real-time image rendering is now feasible with model-based approaches in optoacoustic tomography.
• This advancement opens possibilities for improved diagnostic capabilities in biomedical imaging.